More than 575,000 Americans have end-stage renal disease, and more than 350,000 receive life-saving dialysis treatment. Previous research has demonstrated that patients undergoing hemodialysis, especially those who are older, are at particular risk to have atrial fibrillation. Atrial fibrillation is associated with poor outcomes including high mortality (40% per year). Little is known about fixed and potentially modifiable ris factors for atrial fibrillation in older patients receiving hemodialysis; there is also only scant evidence about these patients' outcomes after their first diagnosis with atrial fibrillation. From already collected Medicare insurance claims and medical records of a large dialysis provider, we propose to assemble a unique dataset that will provide unprecedented detail about these patients health and the health care they receive. In addition, we will have historical Medicare claims predating these patients' time of dialysis initiation by at least 2 years. Using this unique database, we will be in the unusual situation to be able to exclude patients who had atrial fibrillation before they started dialysis, and therefore be able to study incident atrial fibrillaton. We are proposing to identify potentially modifiable risk factors for incident atrial fibrillation, ith particular focus on laboratory measurements, vital signs, dialysis treatment- related factors, and dialysis facility practices. We will also launch a comprehensive assessment of the outcomes of patients once they are first diagnosed with atrial fibrillation, in comparison to similar patients who have not developed this arrhythmia. Outcomes of interest will cover all relevant domains: all-cause and cause-specific mortality; morbidity with focus on thromboembolic and hemorrhagic outcomes; patient-reported health-related quality of life outcomes; and health care utilization and cost. These studies will fill gaping holes in the currently available evidence. The perhaps most innovative aim will focus on deriving a prediction algorithm for near-term risk of new atrial fibrillation using the high-dimensional and extremely granular data in our dataset and novel bioinformatic methods. We will then validate the algorithm in a completely different later time period in data from the same provider as well as in another dataset from a different dialysis provider. If our approach proves successful in identifying patients at the highest near-term risk of incident atrial fibrillation, we are then in the position to test intervention studies to reducethe risk of imminent atrial fibrillation and thus avoid longer term sequelae of this arrhythmia in thes vulnerable patients. Findings from the proposed work have the potential to impact and improve the care that patients with end-stage renal disease receive. Our results may improve the quality of care received and, thus, the outcomes of this vulnerable patient population. The aims and scope of work are in full congruence with the mission of the National Institutes of Diabetes and Digestive and Kidney Diseases, and more specifically the Division of Kidney, Urologic, and Hematologic Diseases, which will consider this application for funding.